A conductor in Tokyo moves his baton, and an orchestra in Cleveland starts to play. A few bars later, a violinist in Berlin joins in. To compensate for a slight delay, the musicians play along with an electronic metronome. The performance is broadcast on high-fidelity speakers and high-definition television. Such a musical experiment would be challenging enough for a television network to pull off; over the Internet, it would be impossible.
That may soon change. Engineers are developing a new type of Internet connection called a dynamic-circuit network that could carry so much data so quickly it might startle even Net surfers in Japan or South Korea. If all goes to plan, the vast data speeds required for such a collaboration may soon be available to all. That might go a long way to solving the problem of how to handle the enormous growth in Internet traffic, which by some estimates is doubling each year.
When a digital photo, YouTube clip or live streaming video is sent over the Internet, the data is first divided into packets, which are routed to the destination and then reassembled. This method tends to break down when large amounts of information are sent along similar pathways at the same time—the different packets are prone to getting hung up on bottlenecks. The result is a Web page that crashes or an online videoconference that sputters and skips. As the Internet starts to stagger under the weight of more and more data, such problems are increasingly common.
Dynamic-circuit network technology would solve the bottleneck problem by opening up a dedicated route from one point to another through which data can be sent without interruption, at speeds thousands of times faster than what's possible now. Just as quickly, the pathway can be closed and space made available for the next connection. The technology would require some new software and fiber-optic links right up to the PC, but otherwise no change in infrastructure.
The potential difference in data speed is staggering. Whereas a PC can now stream two simultaneous video channels, on a dynamic-circuit network it could send 4,000 channels at once. Downloading a high-definition movie over a cable Internet connection of 4.5 megabytes per second would take about three hours. Over a 10-gigabit-per-second dynamic-circuit network, the same movie would download in five seconds.
The technology is being developed by Internet2, a consortium of U.S. universities, companies and scientific and governmental organizations. For now, only a handful of people can use it. It's being tested on the soon-to-be-operational Large Hadron Collider in Europe, where scientists must share massive amounts of data about the collision of subatomic particles. Astronomers are also using it to share data from radio telescopes. And medical researchers plan to use it to collaborate on long-distance examinations and consultations. Internet2 hopes to complete testing in May 2008 and roll the technology out shortly after.
The consortium is working on standards and open-source software designed to make the technology freely available to everybody. But some experts aren't convinced that the Internet needs a new method of sending data. Recent progress in making "packet switching" software more efficient, as well as better optical fiber and faster servers and routers, have convinced them that the old ways work just fine. Even if the new technology isn't essential, says Internet2 CEO Doug Van Houweling, "it's very hard to predict what's going to happen. Every time we add an important new technical capability to the Internet, really smart people figure out how to do things with it that we never imagined."